Evaluation of candidate genes in a genome-wide association study of childhood asthma in Mexicans

Article Outline

• Abstract
• Methods
  • Study design and subject enrollment
  • Clinical evaluation
  • Candidate gene selection
  • Genotyping and quality control
  • Statistical analysis
• Results
• Discussion
• Acknowledgment
• Methods
  • Clinical evaluation
  • Genotyping and quality control
  • Statistical analysis
• Results
• Fig E1. 
• Fig E2. 
• Table E1. 
• Table E2. 
• Table E3. 
• Table E4. 
• Table E5. 
• References
• References
• Copyright

Background

More than 200 asthma candidate genes have been examined in human association studies or identified with knockout mouse approaches. However, many have not been systematically replicated in human populations, especially those containing a large number of tagging single nucleotide polymorphisms (SNPs).

Objective

We comprehensively evaluated the association of previously implicated asthma candidate genes with childhood asthma in a Mexico City population.

Methods

From the literature, we identified candidate genes with at least 1 positive report of association with asthma phenotypes in human subjects or implicated in asthma pathogenesis using knockout mouse experiments. We performed a genome-wide association study in 492 asthmatic children aged 5 to 17 years and both parents using the Illumina HumanHap 550v3 BeadChip. Separate candidate gene analyses were performed for 2933 autosomal SNPs in the 237 selected genes by using the log-linear method with a log-additive risk model.

Results

Sixty-one of the 237 genes had at least 1 SNP with a P value of less than .05 for association with asthma. The 9 most significant results were observed for rs2241715 in the gene encoding TGF-β1 (TGFB1; P = 3.3 × 10⁻⁵), rs13431828 and rs1041973 in the gene encoding IL-1 receptor–like 1 (IL1RL1; P = 2 × 10⁻⁴ and 3.5 × 10⁻⁴), 5 SNPs in the gene encoding dipeptidyl-peptidase 10 (DPP10; P = 1.6 × 10⁻⁴ to 4.5 × 10⁻⁴), and rs17599222 in the gene encoding cytoplasmic FMR1 interacting protein 2 (CYFIP2; P = 4.1 × 10⁻⁴). False discovery rates were less than 0.1 for all 9 SNPs. Multimarker analysis identified TGFB1, IL1RL1, the gene encoding IL-18 receptor 1 (IL18R1), and DPP10 as the genes most significantly associated with asthma.

Conclusions

This comprehensive analysis of literature-based candidate genes suggests that SNPs in several candidate genes, including TGFB1, IL1RL1, IL18R1, and DPP10, might contribute to childhood asthma susceptibility in a Mexican population.

Key words: Allergy, asthma, genetic predisposition to disease, genome-wide association study, single nucleotide polymorphism

Abbreviations used: CYFIP2, Cytoplasmic FMR1 interacting protein 2 gene, DPP10, Dipeptidyl-peptidase 10 gene, ESR1, Estrogen receptor 1 gene, FDR, False discovery rate, GWAS, Genome-wide association study, IL1RL1,, IL-1 receptor–like 1 gene, IL18R1, IL-18 receptor 1 gene, LD, Linkage disequilibrium, ORMDL3, ORM1-like 3 gene, SNP, Single nucleotide polymorphism, TDT, Transmission disequilibrium test, TGFB1, TGF-β1 gene, TRIMM, TRIad Multi-Marker test

Asthma is a complex disease caused by multiple genetic and environmental factors. Two traditional approaches for identification of asthma susceptibility genes are association studies of candidate genes and linkage studies followed by positional cloning. Candidate gene association studies focus on genes plausibly involved in disease pathogenesis or located in a region of linkage for the disease. The majority of proposed asthma susceptibility genes are biologic candidate genes.

In recent years, it has become feasible to interrogate single nucleotide polymorphisms (SNPs) across the genome to identify novel disease-susceptibility genes, an approach known as the genome-wide association study (GWAS). A novel asthma gene, ORM1-like 3 (ORMDL3) has been identified by using the GWAS approach.¹ Incorporating a priori knowledge of disease cause into the statistical analysis and evaluating prioritized SNPs in predefined candidate genes separately can achieve more efficient use of the GWAS data.²

More than 200 asthma candidate genes have been proposed by using human association, positional cloning, and knockout mouse approaches in the past decade.³, ⁴ However, many of them have not been systematically replicated in additional human populations, including genes with a large number of tagging SNPs, such as the genes encoding dipeptidyl-peptidase 10 (DPP10) and estrogen receptor 1 (ESR1). Replication of associations in different populations is crucial for identifying complex disease-susceptibility genes.⁵ A total of 39 candidate genes from the literature were recently examined for association with childhood asthma by using GWAS data in a non-Hispanic white North American population.⁶ In a GWAS of case-parent triads from Mexico City, we comprehensively evaluated associations of more than 200 previously reported candidate genes with childhood asthma.

Back to Article Outline

Methods 

Study design and subject enrollment 

Using the case-parent triad design,⁷, ⁸ we recruited nuclear families consisting of asthmatic children and both their parents. The cases were children aged 4 to 17 years with asthma given diagnoses by a pediatric allergist at the allergy referral clinic of a large public pediatric hospital in central Mexico City (Hospital Infantil de México, Federico Gómez). Children and parents provided blood samples as sources of DNA. A parent, nearly always the mother, completed a questionnaire on the child's symptoms and risk factors for asthma, including parental smoking and residential history.

The protocol was reviewed and approved by the Institutional Review Boards of the Mexican National Institute of Public Health, the Hospital Infantil de México, Federico Gómez, and the US National Institute of Environmental Health Sciences. Parents provided written informed consent for the child's participation. Children also provided informed assent.

Clinical evaluation 

Detailed protocols for clinical evaluation are described in the Methods section of this article's Online Repository at www.jacionline.org. In brief, the diagnosis of asthma was based on clinical symptoms and response to treatment by pediatric allergists at a major referral hospital.⁹, ¹⁰ At a later date, for research purposes, pulmonary function was measured according to American Thoracic Society specifications.¹¹ Atopy was determined by using skin prick tests to a battery of 24 environmental aeroallergens common in Mexico City. Children were considered atopic if the diameter of the skin wheal to at least 1 allergen exceeded 4 mm.

Candidate gene selection 

We included all 118 human asthma candidate genes listed by Ober and Hoffjan³ in 2006. To update the previous review,³ we searched PubMed for the period June 1, 2005, to July 31, 2008, for genes that had at least 1 positive association of SNPs with asthma phenotypes in human subjects. We used the key words “genetic polymorphism” together with “asthma” or “bronchial or airway” or “hyperreactivity or hyperresponsiveness or hypersensitivity.” We also identified genes directly related to asthma phenotypes by using a knockout mouse approach. For the knockout mouse studies, we used the key words “mouse or mice or murine” and “wildtype or knockout” and “disease models, animal” together with “bronchial or airway” and “asthma or inflammation or hyperresponsiveness.” The updated review indentified 156 genes not referenced by Ober and Hoffjan³ for 274 genes.

Among the 274 genes, 19 were not represented on the Illumina HumanHap 550v3 BeadChip (Illumina, San Diego, California; see Table E1 in this article's Online Repository at www.jacionline.org). We also excluded 5 genes on the X chromosome and 4 genes with more than 300 SNPs within the gene region (5 kb upstream of the 5′ end through 1 kb downstream of the 3′ end) on the Illumina 550v3 BeadChip, leaving 246 autosomal genes for analysis (see Table E1). The total number of SNPs was 3326. We selected candidate genes before analysis of genotyping data.

Genotyping and quality control 

Genotyping was done with the Illumina HumanHap 550v3 BeadChip at the University of Washington, Department of Genome Sciences. Standard quality control of GWAS genotyping data was conducted with PLINK¹² or Genotyping Library and Utilities,¹³ as described in the Methods section of this article's Online Repository.

For the candidate gene analysis, more stringent SNP exclusion thresholds were used: minor allele frequency of less than 3% and a Hardy-Weinberg equilibrium P value of less than 1 × 10⁻⁶. Of the 3,326 autosomal SNPs in 246 selected candidate genes, 2933 SNPs in 237 genes (see Table E1) were analyzed in 492 complete case-parent trios.

The SNP coverage for these 237 genes by using the Illumina 550v3 BeadChip is listed in Table E2 in the Online Repository.¹⁴, ¹⁵

Statistical analysis 

We used a log-linear likelihood approach to analyze associations between asthma and the 2933 individual SNPs.⁷ Details regarding the log-linear method are described in the Methods section of this article's Online Repository. The log-linear method was implemented by using the LEM computer program¹⁶ with a 1 df log-additive risk model specified. P values were generated to assess statistical significance, and the relative risk of carrying 1 copy of the risk allele was calculated to assess the direction and magnitude of association under the log-additive model.

To account for multiple comparisons, we calculated the false discovery rate (FDR) q value for each P value for all of the 2933 SNPs analyzed by using the method of Storey and Tibshirani.¹⁷ The FDR is the expected proportion of false-positive results incurred when a particular test result is called significant. However, these corrections will be conservative because the FDR does not take into account the correlation between SNPs. We used the FDR threshold of 0.1 for declaring significance because Van den Oord and Sullivan¹⁸ showed that it achieved a good balance between avoiding false discoveries and detecting true effects.

There is a higher chance of observing SNPs with significant P values for genes with more SNPs. To address this issue, we used a multimarker approach, the TRIad Multi-Marker test (TRIMM), to test the association of asthma with sets of SNPs.¹⁹ This procedure achieves a natural correction for multiple comparisons by treating multiple SNPs as a set and using permutation procedure to evaluate the test significance. In our analysis all SNPs in a gene (5 kb upstream of the 5′ end through 1 kb downstream of the 3′ end) were defined as a set, and a P value was calculated for each gene. For the largest gene on our candidate gene list, DPP10, which spans 1.4 Mb on chromosome 2, the SNPs were divided into 7 sets along the chromosome based on the linkage disequilibrium (LD) structure of the gene (see Table E3 in this article's Online Repository at www.jacionline.org). The P values were estimated for each DPP10 block and the whole gene. We implemented the TRIMM procedure in R (http://www.r-project.org). The R code is available at http://www.niehs.nih.gov/research/atniehs/labs/bb/staff/weinberg.

Back to Article Outline

Results 

Detailed characteristics of the 492 asthmatic children are presented in Table I and described in this article's Results section in the Online Repository at www.jacionline.org. The mean age of cases was 9.0 years (range, 5-17 years). Most had mild as opposed to moderate or severe asthma. Ninety-two percent of cases had at least 1 positive skin test result.

Table I. Demographic and clinical characteristics of the 492 asthmatic children

Clinical characteristics
Age (y), mean ± SD	9.0 ± 2.4
Sex (male)	58.7%
Asthma severity (n = 469)∗
Mild	72.3%
Moderate to severe	27.7%
Asthma medication in the past 12 mo (n = 486)∗	98.2%
FEV1 (% predicted), mean ± SD (n = 371)∗	90.5% ± 16.8%
Skin test positivity (of 24 aeroallergens, n = 445)∗
≥1 Allergen	91.7%
≥5 Allergens	51.5%
Parental smoking (n = 486)∗
Mother smoked during pregnancy	4.8%
Current smoking parent	52.1%

Many of the 2933 analyzed SNPs are in high LD with each other in our Mexican population. Using the LD based SNP-pruning procedure implemented in PLINK (using parameters of window size = 50, number of SNPs to shift at each step = 5, variance inflation factor = 2), we calculated that 1125 SNPs were in approximate linkage equilibrium (variance inflation factor <2) with each other.

Fig 1, A, shows the chromosomal position of all candidate gene SNPs tested for association with asthma and their corresponding significance levels. Fig 1, B, shows the quantile-quantile plot of the P values, indicating the number of observed significant associations exceeding the expected P values under the null hypothesis of no association. Among the 237 asthma candidate genes, 61 had at least 1 SNP with a P value of less than .05 for association with asthma (Table II for SNPs at P < .01 and see Table E4 in this article's Online Repository at www.jacionline.org for SNPs at .01 ≤ P < .05). By using conservative Bonferroni correction for 1125 independent tests (number of SNPs in approximate linkage equilibrium), only rs2241715 in TGF-β1 (TGFB1) met the significance level of 4.4 × 10⁻⁵. However, given that the genes were selected based on prior evidence, Bonferroni correction is overly conservative. Nine SNPs met the FDR q value significance threshold of less than 0.1, including rs2241715 in TGFB1 on chromosome 19 (P = 3.3 × 10⁻⁵, FDR q = 0.059); rs13431828 and rs1041973 in IL-1 receptor−like 1 (IL1RL1) on chromosome 2 (P = 2.0 × 10⁻⁴ for rs13431828 and 3.5 × 10⁻⁴ for rs1041973, FDR q = 0.087 for both); rs980317, rs7421482, rs980316, rs949577, and rs12469474 in DPP10 on chromosome 2 (P = 1.6 × 10⁻⁴ to 4.5 × 10⁻⁴, FDR q = 0.087 for all); and rs17599222 in cytoplasmic FMR1 interacting protein 2 (CYFIP2) on chromosome 5 (P = 4.1 × 10⁻⁴, FDR q = 0.087).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  A, A summary of associations among 2933 autosomal SNPs in 237 candidate genes and childhood asthma in a Mexican population. The x-axis indicates the genomic position of all SNPs divided by chromosome. The y-axis shows the degree of association indicated as −log₁₀ P value. B, Quantile-quantile plot. Distribution of the observed P values for the 2933 SNPs compared with the P values expected under the null hypothesis of no association. Observed −log₁₀ P values are ranked in order on the y-axis and plotted against the corresponding expected −log₁₀ P values on the x-axis. The red line indicates the null distribution.

Table II. Genetic associations between SNPs in candidate genes and childhood asthma in a Mexican population (P < .01)

Chr, Chromosome; MAF, minor allele frequency; RR, relative risk; nonsynon, coding nonsynonymous SNP; UTR, untranslated region.

Phenotypic heterogeneity is a potential factor contributing to failure of replication among different studies. In addition to the primary analysis among the 492 trios, we repeated the log-linear analysis among 378 trios including children with nonmissing skin test and questionnaire data who had positive skin test results and whose mothers did not smoke during pregnancy. The magnitude and direction of the association did not differ appreciably when we analyzed this smaller dataset (see Table E5 in this article's Online Repository at www.jacionline.org).

Results from the multimarker analysis, which corrects for the number of SNPs analyzed in a gene, were consistent with the single-SNP findings (Table III and see Table E2 in this article's Online Repository at www.jacionline.org). The candidate genes that were most significantly associated with asthma were TGFB1 (global P = 2.8 × 10⁻⁴) on chromosome 19q13, IL1RL1 (global P = 2.2 × 10⁻⁴) and the adjacent IL-18 receptor 1 (IL18R1; global P = 9 × 10⁻³) on chromosome 2q12, and DPP10 (global P = 7.8 × 10⁻⁴ for DPP10_block 3 and .05 for the whole gene) on chromosome 2q14.

Table III. Multimarker analysis of associations between candidate genes and childhood asthma in a Mexican population

IL1RL1 is adjacent to IL18R1, located 12 Mb upstream of DPP10 on chromosome 2. Fig E1 (available in this article's Online Repository at www.jacionline.org) shows the pairwise LD (r²) among IL1RL1, IL18R1, and DPP10 SNPs with P values of less than .05 for association with asthma. IL1RL1 and IL18R1 resided in an LD block. The 2 IL1RL1 SNPs, rs13431828 and rs1041973, that were significantly associated with asthma at an FDR q value of less than 0.1 are in moderate LD (r² = 0.46) with each other. These 2 SNPs are potentially functional. The SNP rs13431828 is located in the 5′ untranslated region of IL1RL1, and rs1041973 is a coding nonsynonymous SNP (Glu/Ala) in exon 2. Three additional tightly linked coding nonsynonymous IL1RL1 SNPs, rs10204137, rs10192157, and rs10206753 (r² = 0.97-1), also showed moderate associations with asthma (P = .013 for all 3 SNPs). The 5 DPP10 SNPs, rs980317, rs7421482, rs980316, rs949577, and rs12469474, that were significantly associated with asthma at an FDR q value of less than 0.1 are in moderate to tight LD (r² = 0.39-0.93) with each other and located within the LD block DPP10_block 3 (see Fig E1 and Table E3).

Back to Article Outline

Discussion 

We comprehensively evaluated the association of previously reported asthma genes with childhood asthma in Mexico City within the context of a genome-wide association genotyping platform. Candidate genes were identified from a systematic literature review completed before analysis of the genotyping data. Single-SNP analyses showed that SNPs in TGFB1, DPP10, IL1RL1, and CYFIP2 were significantly associated with childhood asthma in a Mexican population after correction for multiple comparisons by using an FDR approach (FDR q value < 0.1). Our multimarker analysis accounted for gene-wide multiple comparisons by generating a global P value for all SNPs in a region, and these results confirmed that several genes, including TGFB1, DPP10, and IL1RL1, are related to childhood asthma susceptibility.

Compared with traditional candidate gene and linkage studies, the GWAS approach has the advantage of interrogating SNPs across the whole genome to identify novel disease-susceptibility genes unrestrained by prior knowledge. However, questions regarding how to make optimal use of the GWAS data remain unanswered. Li et al² have shown that preselecting SNPs from candidate genes and analyzing this prioritized subset of SNPs separately can improve the power of detecting a disease-susceptibility locus in GWAS.

Many candidate genes have been studied for asthma.³, ⁴ A candidate gene association study usually examines only a relatively small number of SNPs in few selected genes. Many of the published asthma candidate genes, especially large genes with many tagging SNPs, such as DPP10, have not been comprehensively evaluated in additional human populations. Thirty-nine candidate genes were recently evaluated for associations with childhood asthma by using GWAS data from a non-Hispanic white North American population.⁶ We examined a much larger number of candidate genes in a population that has not been well studied.

TGF-β1 is a multifunctional cytokine that might influence asthma by modulating allergic airway inflammation and airway remodeling. TGFB1 is one of the most replicated asthma candidate genes, and SNPs in TGFB1 have been associated with asthma phenotypes in approximately 10 published studies.²⁰ We previously reported that 3 of 5 genotyped TGFB1 SNPs, rs1800469 (C-509 T, a promoter SNP), rs1982073 (T869C, a nonsynonymous SNP), and rs7258445 (an intronic SNP), were associated with asthma in the Mexican population.²¹ In the present analysis we examined 3 additional TGFB1 SNPs, rs2241715, rs4803455, and rs8110090. Fig E2 (available in this article's Online Repository at www.jacionline.org) shows the pairwise LD (r²) among the 8 TGFB1 SNPs that have been examined in our study population to date. The SNP rs2241715, which was significantly associated with asthma in the present analysis, was in moderate to high LD (r² = 0.5-0.95) with the 3 asthma-associated SNPs reported in our previous article.²¹ Two asthma-associated SNPs, rs1800469 and rs1982073, are functional. Rs1800469, also referred to as C-509 T, is located in the promoter region, and this SNP can influence TGF-β1 function, promoter activity, and circulating TGF-β1 levels.²¹ rs1982073, also referred to as T869C, is a nonsynonymous SNP, and the T-to-C substitution leads to an amino acid change from leucine to proline in the signal peptide resulting in increased secretion of TGF-β1 in vitro and increased circulating TGF-β1 concentration.²¹

IL1RL1 is adjacent to IL18R1 and located in an IL-1 (IL1) receptor gene cluster on chromosome 2q12.²² Gene products of IL1RL1 and IL18R1 both belong to the IL-1 receptor family, whose members mediate the signal transduction of IL-1 cytokines during inflammation and host defense.²³ IL-1RL1 binds IL-33 and plays important roles in regulation of T_H2 cell−mediated allergic airway inflammation²⁴, ²⁵ and eosinophil-mediated inflammation.²⁶ Serum levels of IL-1RL1 are increased in atopic asthmatic patients during acute exacerbations.²⁷ IL18R1 encodes the α chain of the IL-18 receptor.²⁸ The IL-18 receptor binds IL-18 and enhances T_H1 cell-driven immune responses in synergy with IL-12.²⁸ IL-18 can also induce the development of T_H2 cells and stimulate T_H2 cytokine release and plays a complicated role in atopic asthma, depending on its immunologic environment.²⁸

SNPs in IL1RL1 and IL18 have been associated with asthma-related phenotypes in only 3 previous studies conducted in several European populations and 1 Korean population.²⁹, ³⁰, ³¹ IL1RL1 and IL18R1 are located together in an LD block in Europeans²⁹, ³⁰ and Mexicans. We examined 11 SNPs in IL1RL1 and 9 SNPs in IL18R1. Eleven of the 20 SNPs were associated with asthma in the Mexican population (P < .01 for 6 SNPs and .01 ≤ P < .05 for 5 SNPs). There is little overlap between the SNPs genotyped across studies.²⁹, ³⁰, ³¹ Two (rs1041973 and rs10206753) of the 4 coding nonsynonymous IL1RL1 SNPs associated with asthma in our Mexican population were also examined in a Dutch population, where they showed no associations.²⁹ An intronic IL1RL1 SNP, rs1420101, or its tightly linked SNP, rs950880 (r² = 0.96 in European HapMap samples), has been significantly associated with blood eosinophil counts and asthma in European and Korean populations,³¹ although not in our Mexican population. The rs1420094 SNP in IL18R1 was significantly associated with atopic asthma in Europeans³⁰ and our Mexican population.

DPP10 was identified as an asthma candidate gene by means of positional cloning,³² but its definitive function is still unclear. DPP10 is a member of the dipeptidyl peptidase family that can remove N-terminal dipeptides from chemokines and cytokines and thus might modify their functional activities.³², ³³ Alternative transcriptional spliced variants of DPP10 are expressed in many tissues, including airways (trachea), and are abundant in T cells.³² SNPs in an LD island across the first 60-kb region of DPP10 intron 1 were associated with asthma in British and German populations.³², ³⁴ Of note, only SNPs in the first 200 kb of the DPP10 genomic DNA were examined for association with asthma-related phenotypes in the original report and the study of Blakey et al.³², ³⁴ A previous examination of DPP10 within a GWAS evaluated 252 SNPs and found that 25 SNPs provided P values of less than .05 for association with asthma in a non-Hispanic white North American population (smallest P = .001).⁶ Among the 253 SNPs we studied, 36 SNPs spreading over a 900-kb genomic region encompassing intron 1 to intron 3 of DPP10 all produced P values of less than .05 for association with asthma in the Mexican population. To our knowledge, no functional DPP10 SNPs have been reported yet. Allen et al³² identified several alternative splicing sites located in an 850-kb region across exon 1, intron 1, and exon 2, which can lead to the production of membrane-bound and other isoforms of DPP10. Polymorphisms in regulatory elements resulting in alternative splicing of DPP10 might explain effects on asthma susceptibility from this region.³²

ORMDL3 was the first asthma candidate gene identified using the GWAS approach.¹ We previously examined rs4378650 in ORMDL3 and rs7216389 in the neighboring gasdermin B (GSDMB or GSDML) in 615 nuclear families.³⁵ rs7216389 in GSDML was also on the Illumina 550 K array used in the present analysis. Although rs4378650 in ORMDL3 was not on the Illumina 550 K array, it can be tagged by rs7216389 (r² = 0.92) in Mexicans.³⁵ The results for rs7216389 from our 2 analyses were consistent (relative risk, 1.20; 95% CI, 1.01-1.43; P = .043 in the previous report with 615 families; relative risk, 1.22; 95% CI, 1.01-1.49; P = .042 in the present analysis of 492 trios; a log-additive risk model with C as the reference allele specified for both analyses).³⁵

Our study has several strengths. The triad design and analysis protects against population stratification, a potential source of bias in an admixed population, such as the Mexican population.⁷ The demographic and clinical characteristics of our asthmatic children are well characterized. Our asthma cases were diagnosed by pediatric allergists at a pediatric allergy specialty clinic of a large public referral hospital. Consultation with this pediatric allergy clinic is a tertiary referral in Mexico, and thus the children in our study had already been seen by a generalist and a pediatrician over time for recurrent asthma symptoms. Diagnoses were made on clinical grounds according to previous guidelines.⁹ We did not test for bronchial hyperreactivity. However, a physician's diagnosis of asthma is a valid outcome compared with objective measurements.³⁶ We had objective data on atopy; skin prick tests revealed the vast majority of these children with asthma (92%) to have positive results to common environmental aeroallergens. Thus all findings might apply primarily to atopic asthma.

We comprehensively evaluated the relationship among SNPs in 237 previously published candidate genes and childhood asthma within the context of a GWAS.³⁷ Our single-SNP and multimarker analysis results suggest that SNPs in multiple genes, including TGFB1, IL1RL1, IL18R1, and DPP10, might contribute to childhood asthma susceptibility in a Mexican population.

Back to Article Outline

We thank the children and parents who participated in this study; Dr Deborah Nickerson and Joshua Smith, University of Washington, for their genotyping services; Kevin Jacobs, National Cancer Institute, for technical assistance with the Genotyping Library and Utilities software; Drs Douglas Bell, Xuting Wang, and Lauranell Burch, National Institute of Environmental Health Sciences; Dr Patrick Sullivan, University of North Carolina at Chapel Hill, for bioinformatics support; Stephanie Holmgren, National Institute of Environmental Health Sciences, for reference services; and Dr Stephan Chanock, National Cancer Institute, for determination of short tandem repeats for parentage testing.

Back to Article Outline

Methods 

Clinical evaluation 

The diagnosis of asthma was based on clinical symptoms and response to treatment by pediatric allergists at a major referral hospital.^E1 The severity of asthma was rated by a pediatric allergist according to symptoms in the Global Initiative on Asthma schema as mild (intermittent or persistent), moderate, or severe.^E2 At a later date, for research purposes, pulmonary function was measured by using the EasyOne spirometer (ndd Medical Technologies, Andover, Mass), according to American Thoracic Society specifications.^E3 The best test of 3 technically acceptable tests was selected. Spirometric prediction equations from a Mexico City childhood population were used to calculate the percent predicted FEV₁.^E4 Children were asked to hold asthma medications on the morning of the test.

Atopy was determined by using skin prick tests to common environmental aeroallergens. A battery of 24 aeroallergens (IPI ASAC, Mexico) common in Mexico City was used: Aspergillus fumigatus, Alternaria species, Mucor species, Blattella germanica, Periplaneta americana, Penicillium species, cat, dog, horse, Dermatophagoides species (both pteronyssinus and farinae), Ambrosia species, Artemisa ludoviciana, Cynodon dactylon, Chenopodium album, Quercus robur, Fraxinus species, Helianthus annus, Ligustrum vulgare, Lolium perenne, Plantago lanceolata, Rumex crispus, Schinus molle, Salsola species, and Phleum pratense. Histamine was used as a positive control, and glycerin was used as a negative control. The test was considered valid if the reaction to histamine was 6 mm or greater according to the grading of the skin prick test, as recommended by Aas and Belin.^E5, ^E6 Children were considered atopic if the diameter of the skin wheal to at least 1 allergen exceeded 4 mm.

Genotyping and quality control 

DNA was extracted from peripheral blood lymphocytes by using Gentra Puregene kits (Gentra System, Minneapolis, Minn). A total of 498 complete case-parent trios with previously confirmed parentage and sufficient amounts of DNA were genotyped for 561,466 SNPs by using the Illumina HumanHap 550v3 BeadChip at the University of Washington, Department of Genome Sciences. Genotypes were determined by using Illumina's BeadStudio Genotyping Module, according to the recommended conditions. There were 1,491 study subjects successfully genotyped, with a genotype call rate exceeding 95% and an average call rate of 99.7%. Three trios were excluded because of a low call rate of 1 family member.

Quality control analyses for the 561,466 SNPs in the GWAS scan were conducted by using PLINK (http://pngu.mgh.harvard.edu/purcell/plink),^E7 unless otherwise stated. SNPs were excluded because of poor chromosomal mapping (n = 173), missing rate of greater than 5% (n = 4125), minor allele frequency (MAF) of less than 1% (n = 16,949), a Hardy-Weinberg equilibrium P value of less than 1 × 10⁻¹⁰ (n = 557), Mendelian errors in more than 2 families (n = 4,945), and heterozygous genotype calls for chromosome X SNPs in more than 1 male subject (n = 380). SNPs with 1 or more discordant genotypes across 14 HapMap replicate samples identified by using the Genotyping Library and Utilities application (http://cgf.nci.nih.gov/development/tooldev.html)^E8 were also excluded (n = 921). All SNP exclusions were made sequentially.

Subject-level quality control verified that no subjects had an unusual autosomal homozygosity or an inconsistent sex between genotype and collected phenotype data. Subject-level quality control then assessed subject relatedness to identify unknown intrafamily and interfamily relationships. This identified 2 duplicated trios and 1 trio with first-degree relative parents requiring exclusion. There were 492 complete case-parent trios in the final analysis data set.

For the candidate gene analysis, more stringent SNP exclusion thresholds were used: MAF of less than 3% and a Hardy-Weinberg equilibrium P value of less than 1 × 10⁻⁶. Of the 3,326 autosomal SNPs in 246 selected candidate genes, 2933 SNPs in 237 genes (Table E1) were retained for the statistical analysis after quality control assessment.

Statistical analysis 

We used a log-linear likelihood approach to analyze associations between asthma and the 2933 individual SNPs passing quality control.^E9 The log-linear likelihood-ratio test is a powerful and more flexible generalization of the transmission disequilibrium test (TDT) and has the advantage of providing estimates of the magnitude of associations in addition to the test significance.^E9 Similar to TDT-based methods for the analysis of case-parent data, such as the family-based association test,^E10 the log-linear model tests the same null hypothesis of no within-family relationship between variant and the disease and achieves robustness against genetic population structure through stratification on the possible parental mating types.^E9, ^E11

The log-linear method was implemented by using the LEM computer program^E12 with a 1 df log-additive risk model specified. When missing parental genotypes occur, the log-linear method uses the expectation-maximization algorithm to infer the missing genotypes, allowing incomplete trios to contribute information and minimizing loss of statistical power.^E13 P values were generated to assess statistical significance, and the relative risk of carrying 1 copy of the risk allele was calculated to assess the direction and magnitude of association under the log-additive model. We also calculated P values for associations between asthma and individual SNPs by using the TDT method in PLINK. As expected, P values were very close using the 2 methods.

Back to Article Outline

Results 

Characteristics of the 492 asthmatic children with genotyping data are presented in Table I. The mean age of cases was 9.0 years (range, 5-17 years). Most had mild (72.3%) as opposed to moderate or severe (27.7%) asthma. Nearly all cases (98.2%) had used medication for asthma in the past 12 months. Wheezing in the past 12 months was reported by 90.1% of the cases, and chronic dry cough was reported by 65.4%. For 73.8% of the cases, asthma symptoms had interfered with daily activities or school attendance in the past 12 months. Among cases with spirometric data, the mean FEV₁ percent predicted was 90.5% (SD = 16.8%). Ninety-two percent of cases had at least 1 positive skin test result. The highest rates of skin test positivity were seen for dust mite (69.7%) and cockroach (43.2%). Only 4.8% of mothers reported smoking during pregnancy, but 51.2% of the children had a parent who currently smoked.

Back to Article Outline

Fig E1. 

• 
  • View Large Image
  • Download to PowerPoint

• Pairwise LD (r²) among IL1RL1, IL18R1, and DPP10 SNPs associated with asthma in a Mexican population (P < .05). SNPs with P values of less than .01 are indicated in green.

Back to Article Outline

Fig E2. 

• 
  • View Large Image
  • Download to PowerPoint

• Pairwise LD (r²) among 8 TGFB1 SNPs that have been examined in our study population to date.^E14 SNPs significantly associated with asthma are indicated in green.

Back to Article Outline

Table E1. 

Candidate genes identified through literature review in the present study

Back to Article Outline

Table E2. 

Multimarker analysis of associations between 237 candidate genes and childhood asthma and SNP coverage for these genes determined by using the Illumina 550v3 BeadChip

Back to Article Outline

Table E3. 

Seven sets of DPP10 SNPs defined based on the LD structure of the gene

Back to Article Outline

Table E4. 

Genetic associations between SNPs in candidate genes and childhood asthma in a Mexican population (.01 ≤ P < .05)

Chr, Chromosome; MAF, minor allele frequency; RR, relative risk; nonsynon, coding nonsynonymous; synon, coding synonymous; UTR, untranslated region.

Back to Article Outline

Table E5. 

Genetic associations between SNPs in candidate genes and childhood asthma in a Mexican population (P < .01)

Chr, Chromosome; RR, relative risk.

Back to Article Outline

References 

1. Moffatt MF, Kabesch M, Liang L, Dixon AL, Strachan D, Heath S, et al. Genetic variants regulating ORMDL3 expression contribute to the risk of childhood asthma. Nature. 2007;448:470–473
   • View In Article
2. Li C, Li M, Lange EM, Watanabe RM. Prioritized subset analysis: improving power in genome-wide association studies. Hum Hered. 2008;65:129–141
   • View In Article
   • CrossRef
3. Ober C, Hoffjan S. Asthma genetics 2006: the long and winding road to gene discovery. Genes Immun. 2006;7:95–100
   • View In Article
   • MEDLINE
   • CrossRef
4. Zhang J, Pare PD, Sandford AJ. Recent advances in asthma genetics. Respir Res. 2008;9:4
   • View In Article
   • CrossRef
5. Chanock SJ, Manolio T, Boehnke M, Boerwinkle E, Hunter DJ, Thomas G, et al. Replicating genotype-phenotype associations. Nature. 2007;447:655–660
   • View In Article
   • CrossRef
6. Rogers AJ, Raby BA, Lasky-Su JA, Murphy A, Lazarus R, Klanderman BJ, et al. Assessing the reproducibility of asthma candidate gene associations using genome-wide data. Am J Respir Crit Care Med. 2009;179:1084–1090
   • View In Article
   • CrossRef
7. Weinberg CR, Wilcox AJ, Lie RT. A log-linear approach to case-parent-triad data: assessing effects of disease genes that act either directly or through maternal effects and that may be subject to parental imprinting. Am J Hum Genet. 1998;62:969–978
   • View In Article
   • MEDLINE
   • CrossRef
8. Wilcox AJ, Weinberg CR, Lie RT. Distinguishing the effects of maternal and offspring genes through studies of “case-parent triads”. Am J Epidemiol. 1998;148:893–901
   • View In Article
   • MEDLINE
9. BTS/SIGN. British guideline on the management of asthma. Thorax. 2003;58(suppl 1):i1–i94
   • View In Article
10. National Heart, Lung, and Blood Institute. Pocket guide for asthma management and prevention: Global Initiative for Asthma. Bethesda: National Institutes of Health; 1998;
    • View In Article
11. American Thoracic Society. Standardization of spirometry, 1994 update. Am J Respir Crit Care Med. 1995;152:1107–1136
    • View In Article
    • MEDLINE
12. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81:559–575
    • View In Article
    • CrossRef
13. Thomas G, Jacobs KB, Yeager M, Kraft P, Wacholder S, Orr N, et al. Multiple loci identified in a genome-wide association study of prostate cancer. Nat Genet. 2008;40:310–315
    • View In Article
    • CrossRef
14. Barrett JC, Cardon LR. Evaluating coverage of genome-wide association studies. Nat Genet. 2006;38:659–662
    • View In Article
    • MEDLINE
    • CrossRef
15. Li M, Li C, Guan W. Evaluation of coverage variation of SNP chips for genome-wide association studies. Eur J Hum Genet. 2008;16:635–643
    • View In Article
    • CrossRef
16. van Den Oord EJ, Vermunt JK. Testing for linkage disequilibrium, maternal effects, and imprinting with (In)complete case-parent triads, by use of the computer program LEM. Am J Hum Genet. 2000;66:335–338
    • View In Article
    • MEDLINE
    • CrossRef
17. Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci U S A. 2003;100:9440–9445
    • View In Article
    • MEDLINE
    • CrossRef
18. van den Oord EJ, Sullivan PF. False discoveries and models for gene discovery. Trends Genet. 2003;19:537–542
    • View In Article
    • MEDLINE
    • CrossRef
19. Shi M, Umbach DM, Weinberg CR. Identification of risk-related haplotypes with the use of multiple SNPs from nuclear families. Am J Hum Genet. 2007;81:53–66
    • View In Article
    • MEDLINE
    • CrossRef
20. Vercelli D. Discovering susceptibility genes for asthma and allergy. Nat Rev Immunol. 2008;8:169–182
    • View In Article
21. Li H, Romieu I, Wu H, Sienra-Monge JJ, Ramirez-Aguilar M, del Rio-Navarro BE, et al. Genetic polymorphisms in transforming growth factor beta-1 (TGFB1) and childhood asthma and atopy. Hum Genet. 2007;121:529–538
    • View In Article
    • MEDLINE
    • CrossRef
22. Dale M, Nicklin MJ. Interleukin-1 receptor cluster: gene organization of IL1R2, IL1R1, IL1RL2 (IL-1Rrp2), IL1RL1 (T1/ST2), and IL18R1 (IL-1Rrp) on human chromosome 2q. Genomics. 1999;57:177–179
    • View In Article
    • MEDLINE
    • CrossRef
23. Dunne A, O'Neill LA. The interleukin-1 receptor/Toll-like receptor superfamily: signal transduction during inflammation and host defense. Sci STKE. 2003;2003:re3
    • View In Article
24. Schmitz J, Owyang A, Oldham E, Song Y, Murphy E, McClanahan TK, et al. IL-33, an interleukin-1-like cytokine that signals via the IL-1 receptor-related protein ST2 and induces T helper type 2-associated cytokines. Immunity. 2005;23:479–490
    • View In Article
    • MEDLINE
    • CrossRef
25. Mangan NE, Dasvarma A, McKenzie AN, Fallon PG. T1/ST2 expression on Th2 cells negatively regulates allergic pulmonary inflammation. Eur J Immunol. 2007;37:1302–1312
    • View In Article
    • MEDLINE
    • CrossRef
26. Cherry WB, Yoon J, Bartemes KR, Iijima K, Kita H. A novel IL-1 family cytokine, IL-33, potently activates human eosinophils. J Allergy Clin Immunol. 2008;121:1484–1490
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (667 KB)
    • CrossRef
27. Oshikawa K, Kuroiwa K, Tago K, Iwahana H, Yanagisawa K, Ohno S, et al. Elevated soluble ST2 protein levels in sera of patients with asthma with an acute exacerbation. Am J Respir Crit Care Med. 2001;164:277–281
    • View In Article
    • MEDLINE
28. Nakanishi K, Yoshimoto T, Tsutsui H, Okamura H. Interleukin-18 is a unique cytokine that stimulates both Th1 and Th2 responses depending on its cytokine milieu. Cytokine Growth Factor Rev. 2001;12:53–72
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (218 KB)
    • CrossRef
29. Reijmerink NE, Postma DS, Bruinenberg M, Nolte IM, Meyers DA, Bleecker ER, et al. Association of IL1RL1, IL18R1, and IL18RAP gene cluster polymorphisms with asthma and atopy. J Allergy Clin Immunol. 2008;122:651–654e8
    • View In Article
    • Full Text
    • Full-Text PDF (391 KB)
    • CrossRef
30. Zhu G, Whyte MK, Vestbo J, Carlsen K, Carlsen KH, Lenney W, et al. Interleukin 18 receptor 1 gene polymorphisms are associated with asthma. Eur J Hum Genet. 2008;16:1083–1090
    • View In Article
    • CrossRef
31. Gudbjartsson DF, Bjornsdottir US, Halapi E, Helgadottir A, Sulem P, Jonsdottir GM, et al. Sequence variants affecting eosinophil numbers associate with asthma and myocardial infarction. Nat Genet. 2009;41:342–347
    • View In Article
    • CrossRef
32. Allen M, Heinzmann A, Noguchi E, Abecasis G, Broxholme J, Ponting CP, et al. Positional cloning of a novel gene influencing asthma from chromosome 2q14. Nat Genet. 2003;35:258–263
    • View In Article
    • MEDLINE
    • CrossRef
33. Kere J, Laitinen T. Positionally cloned susceptibility genes in allergy and asthma. Curr Opin Immunol. 2004;16:689–694
    • View In Article
    • MEDLINE
    • CrossRef
34. Blakey JD, Sayers I, Ring S, Strachan D, Hall I. Positionally cloned Asthma susceptibility gene polymorphisms and disease risk in the British 1958 Birth Cohort. Thorax. 2009;64:381–387
    • View In Article
    • CrossRef
35. Wu H, Romieu I, Sienra-Monge JJ, Li H, Del Rio-Navarro BE, London SJ. Genetic variation in ORM1-like 3 (ORMDL3) and gasdermin-like (GSDML) and childhood asthma. Allergy. 2009;64:629–635
    • View In Article
    • CrossRef
36. Jenkins MA, Clarke JR, Carlin JB, Robertson CF, Hopper JL, Dalton MF, et al. Validation of questionnaire and bronchial hyperresponsiveness against respiratory physician assessment in the diagnosis of asthma. Int J Epidemiol. 1996;25:609–616
    • View In Article
    • MEDLINE
    • CrossRef
37. Hancock DB, Romieu I, Shi M, Sienra-Monge J-J, Wu H, Chiu GY, et al. Genome-wide association study implicates chromosome 9q21.31 as a susceptibility locus for asthma in Mexican children. PLoS Genet. 2009;5:e1000623
    • View In Article

Back to Article Outline

References 

1. BTS/SIGN. British guideline on the management of asthma. Thorax. 2003;58(suppl 1):i1–i94
2. National Heart, Lung, and Blood Institute. Pocket guide for asthma management and prevention: Global Initiative for Asthma. Bethesda: National Institutes of Health; 1998;
3. American Thoracic Society. Standardization of Spirometry, 1994 Update. Am J Respir Crit Care Med. 1995;152:1107–1136
4. Perez-Padilla R, Regalado-Pineda J, Rojas M, Catalan M, Mendoza L, Rojas R, et al. Spirometric function in children of Mexico City compared to Mexican-American children. Pediatr Pulmonol. 2003;35:177–183
5. Aas K, Belin L. Standardization of diagnostic work in allergy. Int Arch Allergy Appl Immunol. 1973;45:57–60
6. Skin tests used in type I allergy testing. Position paper. Sub-committee on skin tests of the European Academy of Allergology and Clinical Immunology. Allergy. 1989;44(suppl):1–59
7. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81:559–575
8. Thomas G, Jacobs KB, Yeager M, Kraft P, Wacholder S, Orr N, et al. Multiple loci identified in a genome-wide association study of prostate cancer. Nat Genet. 2008;40:310–315
9. Weinberg CR, Wilcox AJ, Lie RT. A log-linear approach to case-parent-triad data: assessing effects of disease genes that act either directly or through maternal effects and that may be subject to parental imprinting. Am J Hum Genet. 1998;62:969–978
10. Horvath S, Xu X, Laird NM. The family based association test method: strategies for studying general genotype–phenotype associations. Eur J Hum Genet. 2001;9:301–306
11. Lake SL, Laird NM. Tests of gene-environment interaction for case-parent triads with general environmental exposures. Ann Hum Genet. 2004;68:55–64
12. van Den Oord EJ, Vermunt JK. Testing for linkage disequilibrium, maternal effects, and imprinting with (In)complete case-parent triads, by use of the computer program LEM. Am J Hum Genet. 2000;66:335–338
13. Weinberg CR. Allowing for missing parents in genetic studies of case-parent triads. Am J Hum Genet. 1999;64:1186–1193
14. Li H, Romieu I, Wu H, Sienra-Monge JJ, Ramirez-Aguilar M, del Rio-Navarro BE, et al. Genetic polymorphisms in transforming growth factor beta-1 (TGFB1) and childhood asthma and atopy. Hum Genet. 2007;121:529–538